Back to corpus
technical noteexperiment writeup candidatescore 22

Evo3 Stage 0: RESEARCH — App Fleet Lifecycle Deep Exploration

**Target:** Mega-Cube #13 — "App Fleet Lifecycle" **Date:** 2026-03-10 **Model:** qwen35-cloud (405B) **Exploration Depth:** 3 (recursive) **File Budget:** 30 files max

Open full HTMLDownload HTMLCorpus JSONPaper schema

Full HTML reader

Read the full artifact

Open in new tab

Extracted abstract or opening context

**Target:** Mega-Cube #13 — "App Fleet Lifecycle" **Date:** 2026-03-10 **Model:** qwen35-cloud (405B) **Exploration Depth:** 3 (recursive) **File Budget:** 30 files max ### Candidate 1: M-2 EW Graduation | File | Lines | Key Classes/Functions | Cross-References | |------|-------|----------------------|------------------| | `engine.py` | 400 | `EvolutionWorld.evolve_step()`, `EvolutionWorld.evolve_all()` | graduation.py:28, feedback.py:30 | | `graduation.py` | 234 | `evaluate_graduation()`, `speciate()`, `enter_optimization()`, `enter_dormancy()`, `apply_graduation()` | genome.py:18, feedback.py:31 | | `genome.py` | 241 | `AppGenome`, `Milestone`, `MutationOperator`, `DEFAULT_MILESTONES` | — | | `feedback.py` | 334 | `FeedbackState`, `FeedbackCollector`, `compute_grounded_fitness()`, `compute_metabolism()` | — | **Key Facts:** - Graduation triggers at `fitness >= 0.85` (graduation.py:23) - 3 outcomes: SPECIATION (momentum >0.01/day), OPTIMIZATION (stable), DORMANCY (metabolism <0.15) - Speciation forks genome: stable (RELEASED) + variant (ACTIVE with new milestones) - Optimization locks out G-techniques, keeps R/D only (graduation.py:142-149) - Dormancy preserves capabilities in gene pool for crossover (graduation.py:160) - Feedback has 5 weighted channels with temporal decay (feedback.py:30-36) ### Candidate 2: A-8 Wave 3 | File | Lines | Key Classes/Functions | Cross-References | |------|-------|----------------------|------------------| | `fleet.py` | 329 | `FLEET_3_APPS`, `create_fleet3_genome()`, `initialize_fleet3()`, `run_fleet3_evolution()` | genome.py:18, engine.py:21 | | `population.py` | 453 | `PopulationManager`, `crossover_capabilities()`, `bootstrap_fleet()` | genome.py:24 | **Key Facts:** - 16 Wave 3 apps defined (fleet.py:28-173) - All apps start with `testflight_live=1.0` (already on TestFlight) - Category-specific mutation operator boosts (fleet.py:176-185) - Spore is most mature: `core_feature=0.7`, others at 0.2-0.3 - Categories: language (5), movement (3), builder (4), voice (2), coffee (1), garden (1)

Promotion decision

What has to happen next

Attach run IDs, datasets, metrics, and reproduction commands.

Why this is not always a full paper yet

Corpus pages are public-safe readers for discovered workspace artifacts. They are not automatically final papers. A corpus item becomes a polished paper only after the editable source, evidence checkpoints, references, figures, render path, and release status are attached through the paper schema.